diff --git a/Course%203%20-%20TensorFlow%20Datasets/Week%201/horses-or-humans.ipynb b/Course%203%20-%20TensorFlow%20Datasets/Week%201/horses-or-humans.ipynb new file mode 100644 index 00000000..457c0784 --- /dev/null +++ b/Course%203%20-%20TensorFlow%20Datasets/Week%201/horses-or-humans.ipynb @@ -0,0 +1,399 @@ +{ + "nbformat": 4, + "nbformat_minor": 0, + "metadata": { + "colab": { + "name": "datasets.ipynb", + "provenance": [], + "collapsed_sections": [], + "toc_visible": true, + "include_colab_link": true + }, + "kernelspec": { + "name": "python3", + "display_name": "Python 3" + }, + "accelerator": "GPU" + }, + "cells": [ + { + "cell_type": "markdown", + "metadata": { + "id": "view-in-github", + "colab_type": "text" + }, + "source": [ + "\"Open" + ] + }, + { + "cell_type": "code", + "metadata": { + "id": "zX4Kg8DUTKWO", + "colab_type": "code", + "colab": {} + }, + "source": [ + "#@title Licensed under the Apache License, Version 2.0 (the \"License\");\n", + "# you may not use this file except in compliance with the License.\n", + "# You may obtain a copy of the License at\n", + "#\n", + "# https://www.apache.org/licenses/LICENSE-2.0\n", + "#\n", + "# Unless required by applicable law or agreed to in writing, software\n", + "# distributed under the License is distributed on an \"AS IS\" BASIS,\n", + "# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n", + "# See the License for the specific language governing permissions and\n", + "# limitations under the License." + ], + "execution_count": 0, + "outputs": [] + }, + { + "cell_type": "markdown", + "metadata": { + "id": "coY1OkmCnT_8", + "colab_type": "text" + }, + "source": [ + "Good to run this to ensure you are using TF2.x" + ] + }, + { + "cell_type": "code", + "metadata": { + "colab_type": "code", + "id": "ioLbtB3uGKPX", + "outputId": "c7bf0c0f-1939-405c-ba3a-c262e811997b", + "colab": { + "base_uri": "https://localhost:8080/", + "height": 35 + } + }, + "source": [ + "try:\n", + " # %tensorflow_version only exists in Colab.\n", + " %tensorflow_version 2.x\n", + "except Exception:\n", + " pass" + ], + "execution_count": 0, + "outputs": [ + { + "output_type": "stream", + "text": [ + "TensorFlow 2.x selected.\n" + ], + "name": "stdout" + } + ] + }, + { + "cell_type": "code", + "metadata": { + "id": "yUG2oehtYM5N", + "colab_type": "code", + "colab": {} + }, + "source": [ + "# At time of creation, tfds was at version 2.0 which had\n", + "# some bugs in some common datasets. Advise to update to\n", + "# tfds 2.1.0 like this:\n", + "pip install tensorflow_datasets==2.1.0" + ], + "execution_count": 0, + "outputs": [] + }, + { + "cell_type": "code", + "metadata": { + "id": "QGWOsReCW451", + "colab_type": "code", + "colab": {} + }, + "source": [ + "import tensorflow as tf\n", + "import tensorflow_datasets as tfds\n", + "print(tfds.__version__)\n", + "data, info = tfds.load(\"cnn_dailymail\", with_info=True)" + ], + "execution_count": 0, + "outputs": [] + }, + { + "cell_type": "markdown", + "metadata": { + "id": "u2iz4LvVnXzj", + "colab_type": "text" + }, + "source": [ + "#First Example: Simple dataset\n", + "Let's look at how to use a simple dataset like Fashion MNIST" + ] + }, + { + "cell_type": "code", + "metadata": { + "id": "8YSz4XjbQ0J1", + "colab_type": "code", + "colab": {} + }, + "source": [ + "# Install: pip install tensorflow-datasets\n", + "import tensorflow as tf\n", + "import tensorflow_datasets as tfds\n", + "(training_images, training_labels), (test_images, test_labels) = tfds.as_numpy(tfds.load('fashion_mnist',split = ['train', 'test'], batch_size=-1, as_supervised=True))\n", + "\n", + "training_images = training_images / 255.0\n", + "test_images = test_images / 255.0\n", + "\n", + "model = tf.keras.models.Sequential([\n", + " tf.keras.layers.Flatten(input_shape=(28,28,1)),\n", + " tf.keras.layers.Dense(128, activation=tf.nn.relu),\n", + " tf.keras.layers.Dropout(0.2),\n", + " tf.keras.layers.Dense(10, activation=tf.nn.softmax)\n", + "])\n", + "\n", + "model.compile(optimizer='adam',\n", + " loss='sparse_categorical_crossentropy',\n", + " metrics=['accuracy'])\n", + "\n", + "\n", + "model.fit(training_images, training_labels, epochs=5)" + ], + "execution_count": 0, + "outputs": [] + }, + { + "cell_type": "markdown", + "metadata": { + "id": "rt7Gz9ndnc5r", + "colab_type": "text" + }, + "source": [ + "Code to install tensorflow addons if we are going to use any of its\n", + "libraries for augmentation" + ] + }, + { + "cell_type": "code", + "metadata": { + "id": "TL2RU3aiEzG2", + "colab_type": "code", + "colab": {} + }, + "source": [ + "pip install tensorflow-addons" + ], + "execution_count": 0, + "outputs": [] + }, + { + "cell_type": "markdown", + "metadata": { + "id": "b4kEkLdcnkV6", + "colab_type": "text" + }, + "source": [ + "# Second Example: Horses or Humans with Validation\n", + "This will use Horses or Humans with validation to train with the\n", + "data loaded from TFDS" + ] + }, + { + "cell_type": "code", + "metadata": { + "id": "iSq4t32ZHHpt", + "colab_type": "code", + "colab": {} + }, + "source": [ + "import tensorflow as tf\n", + "import tensorflow_datasets as tfds\n", + "import tensorflow_addons as tfa\n", + "\n", + "data = tfds.load('horses_or_humans', split='train', as_supervised=True)\n", + "val_data = tfds.load('horses_or_humans', split='test', as_supervised=True)\n", + "\n", + "def augmentimages(image, label):\n", + " image = tf.cast(image, tf.float32)\n", + " image = (image/255)\n", + " image = tf.image.random_flip_left_right(image)\n", + " image = tfa.image.rotate(image, 40, interpolation='NEAREST')\n", + " return image, label\n", + "\n", + "train = data.map(augmentimages)\n", + "train_batches = train.shuffle(100).batch(32)\n", + "validation_batches = val_data.batch(32)\n", + "\n", + "model = tf.keras.models.Sequential([\n", + " tf.keras.layers.Conv2D(16, (3,3), activation='relu', \n", + " input_shape=(300, 300, 3)),\n", + " tf.keras.layers.MaxPooling2D(2, 2),\n", + " tf.keras.layers.Conv2D(32, (3,3), activation='relu'),\n", + " tf.keras.layers.MaxPooling2D(2,2),\n", + " tf.keras.layers.Conv2D(64, (3,3), activation='relu'),\n", + " tf.keras.layers.MaxPooling2D(2,2),\n", + " tf.keras.layers.Conv2D(64, (3,3), activation='relu'),\n", + " tf.keras.layers.MaxPooling2D(2,2),\n", + " tf.keras.layers.Conv2D(64, (3,3), activation='relu'),\n", + " tf.keras.layers.MaxPooling2D(2,2),\n", + " tf.keras.layers.Flatten(),\n", + " tf.keras.layers.Dense(512, activation='relu'),\n", + " tf.keras.layers.Dense(1, activation='sigmoid')\n", + "])\n", + "\n", + "\n", + "model.compile(optimizer='Adam', loss='binary_crossentropy', metrics=['accuracy'])\n", + "\n", + "history = model.fit(train_batches, epochs=10, validation_data=validation_batches, validation_steps=1)" + ], + "execution_count": 0, + "outputs": [] + }, + { + "cell_type": "markdown", + "metadata": { + "id": "zeJ2C3kznuvq", + "colab_type": "text" + }, + "source": [ + "# Third Example: Cats v Dogs with custom splits\n" + ] + }, + { + "cell_type": "code", + "metadata": { + "id": "Mr4yTXWn_gFv", + "colab_type": "code", + "outputId": "2f02abf3-8a1e-419f-d395-8099c2878418", + "colab": { + "base_uri": "https://localhost:8080/", + "height": 884 + } + }, + "source": [ + "import tensorflow as tf\n", + "import tensorflow_datasets as tfds\n", + "import tensorflow_addons as tfa\n", + "import numpy as np\n", + "\n", + "\n", + "def augmentimages(image, label):\n", + " image = tf.cast(image, tf.float32)\n", + " image = (image/255)\n", + " image = tf.image.resize(image,(300,300))\n", + " return image, label\n", + "\n", + "count_data = tfds.load('cats_vs_dogs',split='train',as_supervised=True)\n", + "train_data = tfds.load('cats_vs_dogs', split='train[:80%]', as_supervised=True)\n", + "validation_data = tfds.load('cats_vs_dogs', split='train[80%:90%]', as_supervised=True)\n", + "test_data = tfds.load('cats_vs_dogs', split='train[-10%:]', as_supervised=True)\n", + "\n", + "\n", + "#count_length = [i for i,_ in enumerate(count_data)][-1] + 1\n", + "#print(count_length)\n", + "\n", + "#train_length = [i for i,_ in enumerate(train_data)][-1] + 1\n", + "#print(train_length)\n", + "\n", + "#validation_length = [i for i,_ in enumerate(validation_data)][-1] + 1\n", + "#print(validation_length)\n", + "\n", + "#test_length = [i for i,_ in enumerate(test_data)][-1] + 1\n", + "#print(test_length)\n", + "\n", + "#augmented_training_data=train_data.map(augmentimages)\n", + "#augmented_validation_data=validation_data.map(augmentimages)\n", + "#train_batches = augmented_training_data.shuffle(1024).batch(32)\n", + "#validation_batches = augmented_validation_data.batch(10)\n", + "\n", + "augmented_training_data = train_data.map(augmentimages)\n", + "train_batches = augmented_training_data.shuffle(1024).batch(32)\n", + "\n", + "model = tf.keras.models.Sequential([\n", + " tf.keras.layers.Conv2D(16, (3,3), activation='relu', \n", + " input_shape=(300, 300, 3)),\n", + " tf.keras.layers.MaxPooling2D(2, 2),\n", + " tf.keras.layers.Conv2D(32, (3,3), activation='relu'),\n", + " tf.keras.layers.MaxPooling2D(2,2),\n", + " tf.keras.layers.Conv2D(64, (3,3), activation='relu'),\n", + " tf.keras.layers.MaxPooling2D(2,2),\n", + " tf.keras.layers.Conv2D(64, (3,3), activation='relu'),\n", + " tf.keras.layers.MaxPooling2D(2,2),\n", + " tf.keras.layers.Conv2D(64, (3,3), activation='relu'),\n", + " tf.keras.layers.MaxPooling2D(2,2),\n", + " tf.keras.layers.Flatten(),\n", + " tf.keras.layers.Dense(512, activation='relu'),\n", + " tf.keras.layers.Dense(1, activation='sigmoid')\n", + "])\n", + "\n", + "\n", + "model.compile(optimizer='Adam',\n", + " loss='binary_crossentropy',\n", + " metrics=['accuracy'])\n", + "\n", + "history = model.fit(train_batches, epochs=25)" + ], + "execution_count": 0, + "outputs": [ + { + "output_type": "stream", + "text": [ + "Epoch 1/25\n", + "582/582 [==============================] - 65s 111ms/step - loss: 0.6373 - accuracy: 0.6249\n", + "Epoch 2/25\n", + "582/582 [==============================] - 59s 101ms/step - loss: 0.5163 - accuracy: 0.7395\n", + "Epoch 3/25\n", + "582/582 [==============================] - 64s 109ms/step - loss: 0.4232 - accuracy: 0.8045\n", + "Epoch 4/25\n", + "582/582 [==============================] - 64s 109ms/step - loss: 0.3512 - accuracy: 0.8434\n", + "Epoch 5/25\n", + "582/582 [==============================] - 58s 100ms/step - loss: 0.2903 - accuracy: 0.8760\n", + "Epoch 6/25\n", + "582/582 [==============================] - 59s 101ms/step - loss: 0.2264 - accuracy: 0.9050\n", + "Epoch 7/25\n", + "582/582 [==============================] - 64s 110ms/step - loss: 0.1766 - accuracy: 0.9286\n", + "Epoch 8/25\n", + "582/582 [==============================] - 59s 101ms/step - loss: 0.1197 - accuracy: 0.9523\n", + "Epoch 9/25\n", + "582/582 [==============================] - 63s 109ms/step - loss: 0.0843 - accuracy: 0.9671\n", + "Epoch 10/25\n", + "582/582 [==============================] - 59s 101ms/step - loss: 0.0633 - accuracy: 0.9761\n", + "Epoch 11/25\n", + "582/582 [==============================] - 59s 101ms/step - loss: 0.0586 - accuracy: 0.9801\n", + "Epoch 12/25\n", + "582/582 [==============================] - 59s 101ms/step - loss: 0.0430 - accuracy: 0.9841\n", + "Epoch 13/25\n", + "582/582 [==============================] - 63s 108ms/step - loss: 0.0388 - accuracy: 0.9879\n", + "Epoch 14/25\n", + "582/582 [==============================] - 63s 109ms/step - loss: 0.0439 - accuracy: 0.9858\n", + "Epoch 15/25\n", + "582/582 [==============================] - 58s 100ms/step - loss: 0.0269 - accuracy: 0.9903\n", + "Epoch 16/25\n", + "582/582 [==============================] - 63s 108ms/step - loss: 0.0373 - accuracy: 0.9881\n", + "Epoch 17/25\n", + "582/582 [==============================] - 59s 101ms/step - loss: 0.0312 - accuracy: 0.9896\n", + "Epoch 18/25\n", + "582/582 [==============================] - 63s 108ms/step - loss: 0.0280 - accuracy: 0.9910\n", + "Epoch 19/25\n", + "582/582 [==============================] - 63s 107ms/step - loss: 0.0275 - accuracy: 0.9918\n", + "Epoch 20/25\n", + "582/582 [==============================] - 58s 99ms/step - loss: 0.0226 - accuracy: 0.9929\n", + "Epoch 21/25\n", + "582/582 [==============================] - 58s 100ms/step - loss: 0.0343 - accuracy: 0.9888\n", + "Epoch 22/25\n", + "582/582 [==============================] - 63s 109ms/step - loss: 0.0253 - accuracy: 0.9913\n", + "Epoch 23/25\n", + "582/582 [==============================] - 63s 108ms/step - loss: 0.0211 - accuracy: 0.9928\n", + "Epoch 24/25\n", + "582/582 [==============================] - 58s 100ms/step - loss: 0.0251 - accuracy: 0.9919\n", + "Epoch 25/25\n", + "582/582 [==============================] - 58s 100ms/step - loss: 0.0315 - accuracy: 0.9922\n" + ], + "name": "stdout" + } + ] + } + ] +} \ No newline at end of file